THE THERMOMETERS. 

 The thermometers used were three in number : 



No. 29110, range to 30 C. 

 41203, -30+10 C. 

 41204, -30+10 C. 



They were divided in fifths of a degree Centigrade. The errors were determined 

 at the Reichsanstalt before starting and again after return. These errors are put 

 down in the following table : 



No. 29110. 



May, 1910 C.-O- 10 Too high 

 10 C. -0-06 

 20 C. -0-06 

 30C.-0-16 



Nov., 1913 C.-O- 12 Too high 

 10C.-0-06 

 20 C. 0-08 

 30 C.-O- 16 



No. 41203. 



No. 41204. 



May, 1910 20 C. 0-04 Too low 



-10 C. 0-02 

 0C. 0-02 

 +10 C. 0-02 



May, 1910 20 C. 0-06 Too low 



-10 C. 0-02 



0C. 0-00 



+ 10 C. 0-04 



Oct., 1913 21 C. 0-02 Too high; Oct., 1913 21 C. 0-02 Too high 



11 C. 0-04 



0C.O-04 



+ 10 C. 0-04 



-irc.o-04 



0C.O-06 

 +10 C. 0-02 



The change in error was assumed to be regular during the interval between the 

 two standardizations, and the applied corrections are calculated on this assumption. 



For measurement of the atmospheric pressure, the standard mercury barometer 

 of each station was generally used, and the appropriate corrections applied. 



For hygrometric measurements, recourse was had in general to wet and dry bulb 

 thermometers, the vapour pressure of the moisture in the air being calculated from 

 the temperatures indicated by the two. 



With this particular form of apparatus, the pendulums are swung at atmospheric 

 pressure and corrections are applied to reduce these to measurements in vacua. Owing 

 to damping by the air, the time of observation of any pendulum cannot be unduly 

 prolonged, as otherwise the amplitude of the final oscillations becomes so small as to 

 vitiate the accuracy of the result. 



PROCEDURE IN MAKING OBSERVATIONS. 



The method of procedure followed is to put down the instants of the first 10 or 20 

 coincidences (estimated to the nearest tenth of a second), to calculate the time of the 

 61st coincidence, to observe this one and the instants of the following 10 or 20 coinci- 

 dences. The mean of the differences between the 1st and the 61st, the 2nd and the 

 62nd, and so on, gives the value of 60 coincidence intervals, from which the time of 

 swing of the pendulum is readily deduced.* Barometer, temperature, hygrometer, 



* One half of these coincidences are with ascending flash and the other half with descending flash. 



